Tool to Perform Software-In-the-Loop through Robot Operating System Mauricio Mauledoux 1, a, Crhistian Segura 1, b, Oscar F.

Transcription

1 Applied Mechanics and Materials Submitted: ISSN: , Vols , pp Accepted: doi: / Online: Trans Tech Publications, Switzerland Tool to Perform Software-In-the-Loop through Robot Operating System Mauricio Mauledoux 1, a, Crhistian Segura 1, b, Oscar F. Aviles 1, c 1 Mechatronics Engineering Program, Universidad Militar Nueva Granada, Bogotá, Colombia. a mauricio.mauledoux@unimilitar.edu.co, b crhistian.segura@unimilitar.edu.co, c oscar.aviles@unimilitar.edu.co Keywords: ROS, Raspberry-Pi, SIL, feedback state. Abstract. This article describes the use of Software-in-the-loop (SIL) and Robot Operating System (ROS) as tools for controller implementation and simulation of discrete-time plants is exposed. For experimental validation a magnetic levitation plant is used, this is modeled using Lagrange obtaining a nonlinear model which is linearized. Thus this model is discretized using a Tustin transformation for subsequent implementation of the control loop. Feedback state variable is implemented as control strategy for experimental validation on a system (Raspberry-Pi / fit-pc, Matlab / PC). We chose to use ROS as it is available for computers running operating systems based on Linux, as used in various embedded systems commercially available com the Fit-PC, Beagle-Board and Raspberry-Pi, ROS occupies low disk space (basic installation), programming is done in C ++ allowing more thorough use of the hardware. For testing three modules (node) implemented; "Reference_node" which is responsible for requesting the user to the desired position and transmit it to the next node, "control_node" is responsible for carrying out checks, which receives as inputs the reference (desired position) and the output of the plant (position current), and which outputs the control signal (u), finally "plant_node" is the node that simulates the behavior of the plant. Introduction This article describes the use of Software-in-the-loop (SIL) and Robot Operating System (ROS) as tools for controller implementation and simulation of discrete-time plants is exposed to SIL has been used for different models implementation as observed in [1, 2]. For this paper is used a magnetic levitation plant as reference system, which is modeled using the Lagrange formalism obtaining a nonlinear model, using the Jacobian is obtained a linear representation [3]. With this new model is performed a discretization using the Tustin transformation getting a system in Z. As a control strategy a discrete feedback state variable is implemented on a system (Raspberry-Pi / fit-pc, Matlab / PC). We chose to use ROS as it is available for computers running operating systems based on Linux, as used in various embedded systems commercially available com the Fit-PC, Beagle-Board and Raspberry-Pi, ROS occupies low disk space (basic installation), programming in C ++ allowing more thorough use of the hardware, examples of this are described in [4,5,6] is done. For testing three modules (node) implemented; "Reference_node" which is responsible for requesting the user to the desired position and transmit it to the next node, "control_node" is responsible for carrying out checks, which receives as inputs the reference (desired position) and the output of the plant (position current), and which outputs the control signal (u), finally "plant_node" is the node that simulates the behavior of the plant. Modeling System Along these sections linearization and discretization of a magnetic levitation is presented along with the implementation of a feedback controller state variables, all embedded in an embedded system, and programmed with the ROS tools [7]. The following is intended to explain the procedure used to obtain a digital model that is implementable linearized in a control loop in discrete time for comparing two methods of control such as feedback on state variables and a PID controller. The All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of Trans Tech Publications, (ID: , Pennsylvania State University, University Park, USA-06/03/16,12:11:38)

2 2392 Mechatronics Engineering and Modern Information Technologies in Industrial Engineering plant used to test modeling of magnetic levitation, see figure. 1, where m is the mass of the sphere, g the gravitational constant, the winding resistance R and inductance L. Fig.1. Physical Model. Mathematical modeling of the system was performed by Newton's second law and Kirchhoff's Law; in order to obtain the differential equations, see equation (1) which allows representing the system behavior. As can be evidenced the mathematical model is not linear, in order to implement linear control methods must be eliminated from the equation component non-linear see equation. (2). Thus these equations can be implemented in a digital computer to perform the simulations. The term that added nonlinearity to the model, refers to the desired position of the sphere ( ). For all the simulation values used for each element are: mass (m = 0.1Kg), gravity (g = 9.8m / s ^ 2), winding resistance (R = 1Ω), coil inductance (L = 0.01H) and EMF (c = 1V). (1) (2) Representation to state variables change to perform calculations matrix form as explained in [9], leaving a system in state space, in equation. (3, 4) is found the model of levitation in this formalism, figure. 2.a, has the model of the plant in Matlab using matrices A, B, C, D, as shown in figure.2.b by applying a step input to the linearized system 10cm. In order to implement a driver must know the characteristics of the plant, such as the location of poles and zeros of the system, these can be obtained by calculating the characteristic polynomial det(si-a) and solving for the roots see equation (4), in figure.2.c is a diagram of poles and zeros where a pole located in the right half plane which makes the system unstable is observed. (4) (a)linearized system. (b) Response to step input 10cm. (c) Zero pole diagram. Fig.2. Magnetic levitator. After knowing the number and position of poles necessary to verify that the system is observable and controllable which is done by calculating the determinant of the controllability and observability matrices, see Eq. (5), hence it has to be both are non-zero, it can be implemented with a controller and an observer to obtain the desired output.

3 Applied Mechanics and Materials Vols (5) Feedback State Variable The For feedback control state variable model was used in state space as proposed in [5], but because the output of the system only allows direct measurement of one of the states, is used an observer which obtain other states, to obtain the values of (K) controller and the observer (L) that were implemented, the method of Ackerman was used. The design parameters where chosen as follow TS 0.7 and 0.5s, the entire system is shown in figure. 3.a, and the control constants in the equation (6), in figure 3.b, the response of the controlled system is observed. (6) ROS (a) Model Controlled by a feedback state. (b) Feedback system at 10cm step input. Fig.3. Feedback State Variable To perform software testing multiple nodes in the loop was implemented and information sharing is called topics, in figure. 4.a is the architecture developed in ROS, ref_node nodes are in charge of asking the user to the desired reference and transmitted by topic "/ref" to receiving further ctrl_node /salt and /obs internally performs the calculation of the error and generates the control signal through the topic /ctrl. The planta_node has implemented the difference equation that generates the output of the system against the entry submitted, but since the matrix C only get a state, it is necessary to implement an observer to estimate the other two states that requires feedback method pos state space, this is done in obs_node. The equations are in the observer (7) where A, B and C are the system matrices, Kc is the observer gain matrix, x and y are the estimated variables and estimated output respectively. Figure. 4.b shows the hierarchy of programs in ROS there you can see that it is necessary to initiate the roscore node, which is responsible for coordinating them all and also creates a server, this allows the other nodes can be launched from other computers on the same network, just pointing to the IP of the PC that launched the roscore. One can show that the other nodes are at the same hierarchical level therefore can be started in any order. (7) (b) Architecture of modules in ROS. (b) Eight nodes and hierarchy in ROS. Fig.4. ROS Architecture.

4 2394 Mechatronics Engineering and Modern Information Technologies in Industrial Engineering Conclusion The use of ROS as a tool for system simulation using the technique of SIL, controllers designed to validate within a discrete control loop, thus plants can be tested in environments without risking critical working machinery, equipment or lives. The use of embedded systems with operating systems that have a low consumption of resources, has the advantage that the implemented algorithms are run efficiently, in the case of the tool allows ROS can ensure signal generation times, and this so you can get a good representation of the continuous discrete signals. ROS has implemented communication between "nodes" that facilitates the implementation of multiple programs that can be used and the protocols established by default or implement own, also stores information transferred allowing for the case of discrete systems controls and see "samples" or past events to implement difference equations. Acknowledgement This work was financially supported by Universidad Militar Nueva Granada through the ING-1571 project titled "Design and construction of robotic land mobile platform for mining exploration ". References [1].M. Hubera, C. Bonsa, D. Mu llera, Exergetic evaluation of solar controller using Software-In-The-Loop method, ScienceDirect ELSEVIER , 2014Demers, S.; Gopalakrishnan, P.; Kant, L., "A Generic Solution to Software-in-the-Loop," Military Communications Conference, MILCOM IEEE, vol., no., pp.1,6, Oct [2].Hassani, K.; Won-Sook Lee, "A software-in-the-loop simulation of an intelligent microsatellite within a virtual environment," Computational Intelligence and Virtual Environments for Measurement Systems and Applications (CIVEMSA), 2013 IEEE International Conference on, vol., no., pp.31,36, July 2013 [3].F. Grognarda, R. Sepulchreb, G. Bastina, Global stabilization of feedforward systems with exponentially unstable Jacobian linearization, [4].I. Mayachita, R. Widyarini, H. R. Sono, A. R. Ibrahim, W. Adiprawita Implementation of Entertaining Robot on ROS Framework ScienceDirect ELSEVIER , 2011 [5].André Araújoa, David Portugala,*, Micael S. Couceiroa,b, Jorge Salesc y Rui P. Rochaa, Desarrollo de un robot móvil compacto integrado en el middleware ROS, , 2014 [6].Fetter Lages, Walter; Ioris, Darlan; Santini, Diego Caberlon, "An Architecture for Controlling the Barrett WAM Robot Using ROS and OROCOS," ISR/Robotik 2014; 41st International Symposium on Robotics; Proceedings of, vol., no., pp.1,8, 2-3 June 2014 [7].DeMarco, K.; West, M.E.; Collins, T.R., "An implementation of ROS on the Yellowfin autonomous underwater vehicle (AUV)," OCEANS 2011, vol., no., pp.1,7, Sept [8].Speers, A; Forooshani, P.M.; Dicke, M.; Jenkin, M., "Lightweight tablet devices for command and control of ROS-enabled robots," Advanced Robotics (ICAR), th International Conference on, vol., no., pp.1,6, Nov [9].K. Ogata, Ingenieria de Control Moderna, Prentice Hall; 3rd edition edition (March 1999).

5 Mechatronics Engineering and Modern Information Technologies in Industrial Engineering / Tool to Perform Software-in-the-Loop through Robot Operating System /